IBM Presence detection Milan Stezka (stezkmil@fel.cvut.cz) - - PowerPoint PPT Presentation

ibm presence detection
SMART_READER_LITE
LIVE PREVIEW

IBM Presence detection Milan Stezka (stezkmil@fel.cvut.cz) - - PowerPoint PPT Presentation

IBM - CVUT Student Research Projects IBM Presence detection Milan Stezka (stezkmil@fel.cvut.cz) Introduction Task: Design and realize an extension to the Smart Kiosk project which will detect presence of a person. Utilize the


slide-1
SLIDE 1

IBM - CVUT Student Research Projects

IBM Presence detection

Milan Stezka (stezkmil@fel.cvut.cz)

slide-2
SLIDE 2

IBM - CVUT Student Research Projects 2

Introduction

  • Task:

– Design and realize an extension to the Smart Kiosk project which will detect presence of a person. – Utilize the OPENCV library and a camera. – Part of the design will be a scenario taking advantage of this information.

slide-3
SLIDE 3

IBM - CVUT Student Research Projects 3

Implementation

  • Used technologies:

– C++ (no virtual machine code bloat and required by OpenCV anyways...) – OpenCV (face detection, recognition and other image processing) – Nokia Qt (GUI) – Windows Sockets (client/server communication) – DirectShow (video overlay drawing)

slide-4
SLIDE 4

IBM - CVUT Student Research Projects 4

Task distribution

  • Server:

– Receives image data from clients – Performs face detection and recognition – Holds user registration data – Sends recognition results to clients

  • Client:

– Displays the results – Shows forms for login, registration, etc.

slide-5
SLIDE 5

IBM - CVUT Student Research Projects 5

Used face recognition in detail

  • Face detection: Viola and Jones (based on Haar

wavelets)

  • Face recognition:

– Eigenfaces – Principal Component Analysis (PCA)

slide-6
SLIDE 6

IBM - CVUT Student Research Projects 6

Does it work?

  • It works quite well :) (for a one semester

long assignment, that is)

– Reasonably stable lighting conditions are required – At least two people (I mean their face images) are required to create working face recognition data